| name | seo-foundations |
| description | Technical backbone for the Career Hub vital loop. Five SEO + website operating standards (technical SEO health / E-E-A-T + linkable-asset link building / programmatic SEO discipline (Schwartz) / Core Web Vitals + perf budgets / operational reliability with SRE primitives). Plus topical-cluster model, intent-mapping, GEO methodology (Princeton KDD 2024), and runbooks library. |
| metadata | {"family":"seo","owner":"seo","last_reviewed":"2026-04-29T00:00:00.000Z","version":"1.0.0","related_skills":["marketing-foundations","seo-audit","seo-page-rules","schema-markup","ai-seo","site-architecture"],"kpis":["Indexation ≥90% of launched-pillar URLs in 30d","Schema validity 100% on templated pages","Cannibalization <5% of pages with sibling ranking same query","Anchor diversity <40% exact-match per target URL","p75 LCP <2.5s, INP <200ms, CLS <0.1 every page type"],"marketing_pillar":4,"seo_standard":"all","funnel_stage":"all","content_class":"all","maturity_stage":"prescriptive","used_by_stages":[1,2,5,6,7,8]} |
SEO Foundations (Career Hub)
Technical backbone for every stage agent in the vital loop. Read first when scoring opportunities, validating templates, gating Stage 8 verification, or running incident response. Anchored in named research — every rule cites its source.
When to Use
- Stage 1 Discovery scoring opportunities (Tier 1 indicators, intent-mapping, pSEO gates).
- Stage 2 Strategy writing briefs (intent + template + topical-cluster placement).
- Stage 5 Brand+Guardrail (GEO citation density per Aggarwal et al.).
- Stage 6 Template (schema rules, internal-linking rules, CWV budgets).
- Stage 7 UI (perf budgets, render strategy).
- Stage 8 UX verification (full standards check pre-ship).
- Incident response (runbooks for indexation/traffic/schema/sitemap/CWV regressions).
Do Not Use When
- Net-new content drafting (use
content/ family).
- Marketing strategy decisions (use
marketing/marketing-foundations).
The Five Standards (one-line each, full content in references)
- Standard A — Technical SEO health: indexation ≥90% in 30d, schema validity 100%, anchor diversity <40% exact-match, cannibalization <5%, crawl health green. YMYL pages (Pillar 3) require highest trust per Google Search Quality Rater Guidelines.
- Standard B — E-E-A-T + linkable-asset link building: AuthorByline coverage 100% on guides + financial tips + role pages, citation density ≥1 per dollar amount (BRAND.md Four-Source Rule), freshness within template SLA, third-party authority targeted. Linkable-asset taxonomy at reference/linkable-asset-taxonomy.md.
- Standard C — Programmatic SEO discipline (Schwartz framework): from Eli Schwartz's Product-Led SEO (2021) — database content + user-generated value + validated demand + the "1,000 useful > 100,000 thin" rule. See reference/pseo-quality-gates.md.
- Standard D — Core Web Vitals + perf budgets: p75 LCP <2.5s, INP <200ms, CLS <0.1. Per-route JS budgets (homepage ≤80KB, tools ≤120KB, role/city/guide ≤60KB, programmatic ≤40KB). RSC-first,
next/image everywhere, single web font, security headers maintained. See reference/cwv-budgets.md.
- Standard E — Operational reliability (SRE primitives): every standard has an SLO, an SLI, and an error budget. Runbooks per common failure type at reference/runbooks/README.md. DORA-style pipeline metrics. See reference/excellence-targets.md.
Topical Authority Model (Pillar-Cluster)
Per HubSpot 2017 + Brian Dean's Topic Cluster architecture. Career Hub already implements this; the model makes it explicit and measurable.
- Pillar page = the broad-topic owner (e.g., role hub for "Bartender jobs"). Internal links DOWN to all clusters.
- Cluster pages = narrow subtopics (how-to-become, interview-questions, resume-examples, salary-by-city, comparison). Each links UP to the pillar.
- Cross-cluster links = sparingly, only when contextually relevant.
- PageRank flow modeled per pillar.
Full cluster topology at reference/topical-cluster-map.md. Each pillar in nextjs-app/docs/PILLARS.md is an SEO topical cluster.
GEO Methodology
Per Aggarwal et al. (Princeton, KDD 2024). The 9 strategies tested in generative engines, ranked by visibility lift:
- citations: +40%
- statistics: +37%
- quotations: +30%
- authoritative tone: +25%
- improved clarity: +20%
- technical terms: +18%
- unique vocabulary: +15%
- fluency + statistics combined: max effect (up to +115% for low-authority sites)
- keyword stuffing: -10% (actively harmful)
brand-lint enforces the top three at template level: ≥1 citation per dollar amount, ≥1 statistic per H2 section, ≥1 quotation per data-rich page. Stage 5 fails any draft missing these.
Full GEO methodology at reference/geo-methodology.md. Already partially in .agents/skills/seo/ai-seo/SKILL.md.
Intent Mapping
Every Strategic Brief tags intent. Intent → page type → schema → CTA pattern at reference/intent-mapping.md. Intent-value multipliers calibrated quarterly from Tier 4 attribution.
Cannibalization Rules
<5% of pages with a sibling ranking for the same primary query. Detection via gsc-mcp.query_search_analytics query × page join. Resolution patterns at reference/cannibalization-rules.md.
Internal Linking Rules
Hub-spoke architecture per pillar. Every page meets:
- min internal links IN per template
- min internal links OUT per template
- anchor diversity <40% exact-match per target URL
- contextual cross-cluster links where appropriate
See reference/internal-linking-rules.md.
Schema Rules
Required JSON-LD per template type (FAQ + Breadcrumb + Article | Occupation | SoftwareApplication | JobPosting). 100% Rich Results Test validity. Full rules at reference/schema-rules.md.
Runbooks
Incident response runbooks at reference/runbooks/README.md:
indexation-regression.mdtraffic-drop.mdschema-breakage.mdsitemap-drift.mdcwv-regression.mdnoindex-accident.mdpseo-thinning.md
Each runbook follows: detect → classify (S1/S2/S3) → contain → fix → blameless postmortem.
Quarterly Weight Tuning
Three weight files versioned in git, tuned monthly via monthly-attribution-review:
